Virtual 3D modeling of thermal power plant turbines based on bilateral filtering algorithm for point cloud data

Abstract

Point cloud segmentation and recognition for virtual 3D wind turbine modeling in thermal plants are disjointed, limiting coverage and elevating mean square error. A design leveraging bilateral filtering of point cloud data for virtual 3D wind turbine modeling is proposed to address this. This approach encompasses data acquisition, preprocessing, and multi-level segmentation to enhance coverage. Bilateral filtering is then applied to calculate the 3D model, with edge correction and verification. Tests reveal that compared to traditional k-means clustering, reduced project space, and hybrid filtering methods, the proposed bilateral filtering algorithm significantly reduces mean square error, maintaining it below 10, demonstrating improved stability, safety, and practical value for thermal plant fan modeling.